Hello! When using ResourceManager, is there a way to make sure that the cleanup method is called when the flow is cancelled from the UI? From some testing I noticed that cleanup wasn't called before the flow was cancelled.

Hi! I’m trying to leverage Kubernetes for Prefect. Our K8s agent is up and is receiving flow runs properly. However, the K8s job creation is failing due to

Error creating: pods "prefect-job-138dbd9d-v6rq5" is forbidden: failed quota: xxxxx-request-quota: must specify requests.cpu,requests.memory

although I specified the request limit with

env:
.....
- name: JOB_MEM_REQUEST
  value: "2048Mi"
- name: JOB_MEM_LIMIT
  value: "4096Mi"
- name: JOB_CPU_REQUEST
  value: "100m"
- name: JOB_CPU_LIMIT
  value: "200m"

This is the Prefect source code if I understand correctly and I’m using

prefect:0.15.5-python3.6

. Any suggestions?

Hi! The first ETL flow is success. But the next scheduled dont run. Please help me

Hi everyone, I have a flow that spins up

n

different jobs in a k8s cluster to perform some computationally heavy tasks. Sometimes when I cancel the flow or it fails for whatever reason, I then need to manually clean up the

n

jobs in the k8s cluster. To avoid this manual work, I am trying to create a state handler to deal with that. The handler works for the failed state, but does not with cancelling. I have tried both

Cancelling

and

Cancelled

. For ilustration purpouses, there is an example below:

import os
import uuid
import time
import prefect
from prefect import Task, Parameter
from prefect.executors.dask import LocalDaskExecutor
from prefect.utilities.edges import unmapped


def clean_up(task, old_state, new_state):
    logger = prefect.context.get("logger")
    task_full_name = prefect.context.get("task_full_name")
    if isinstance(new_state, prefect.engine.state.Failed):
        <http://logger.info|logger.info>(
            f"cleaning up job {task.run_id}-{task_full_name} due to new state being {new_state}"
        )
    elif isinstance(new_state, prefect.engine.state.Cancelled):
        <http://logger.info|logger.info>(
            f"cleaning up job {task.run_id}-{task_full_name} due to new state being {new_state}"
        )
    elif isinstance(new_state, prefect.engine.state.Cancelling):
        <http://logger.info|logger.info>(
            f"cleaning up job {task.run_id}-{task_full_name} due to new state being {new_state}"
        )


class DoSomeComputation(Task):
    def __init__(self, **kwargs):
        self.run_id = str(uuid.uuid4())[:8]
        super().__init__(**kwargs)

    def run(self, input, fail=True):
        logger = prefect.context.get("logger")
        <http://logger.info|logger.info>(f"Started some computation task with input {input}")
        for i in range(0, 200):
            time.sleep(1)
        if fail:
            raise Exception("Something terrible happened")
        <http://logger.info|logger.info>("Sucessful completed some computation task")


if __name__ == "__main__":
    project_name = "test-flows"
    flow_name = "test-cleanup"
    with Flow(...) as flow:
        fail = Parameter("fail")
        do_some_comp = DoSomeComputation(
            state_handlers=[clean_up], task_run_name="{task_name}-job-name={input}"
        )
        inputs = ["input1", "input2", "input3"]
        do_some_comp.map(inputs, unmapped(fail))

    flow.executor = LocalDaskExecutor(num_workers=20)
    flow.register(project_name=project_name)

The "cleaning up job..." log never gets triggered when I cancel the job. Anyone has an ideia or also experienced this? PS: I am using the

KubernetesRun

flow config

Hello everyone! How do I get task timeout value from GraphQL? When I query interactive API I get an error. My request:

query DMS{
  flow (
    where: {
      name: {_eq: "Daily metrics summary"}, archived: {_eq: false}
    }
  ) { 
      id, version,
      tasks {
        id, name,
        task_runs(order_by: {start_time: desc}, limit: 1){
          start_time, end_time
        },
      # Here I'm getting error (see below)
      timeout
    }}
}

Hi, folks. I need to run “create_container” after successfully finished any task in ifelse statement. How to implement upstream dependency for create_container and ifelse statement?

@task(name="Download video", log_stdout=True)
def download_video(task_params, destination_dir='./data'):
    remote_file_path = f"{task_params['storage']['source']}{task_params['video']['filepath']}"
    destination_dir = Path(destination_dir)
    destination_dir.mkdir(exist_ok=True)
    destination_path = f"{destination_dir / Path(remote_file_path).name}"
    file_download(remote_file_path, destination_path)


@task(name="Fetch previous result", log_stdout=True)
def fetch_prev_result(task):
    prev_result = task['prev_result']
    db_name = prev_result['db_name']
    collection_name = prev_result['collection_name']
    out_path = f"{db_name}_{collection_name}.csv"
    mongo_fetch(mongo_client, db_name, collection_name, out_path=out_path)


is_detection = check_equal(task_type, 1)
ifelse(is_detection, download_video(task_params, destination_dir='./videos'), fetch_prev_result(task))
command = '''python -c "from prefect import Flow; f = Flow('empty'); f.run()"'''
container_name = StringFormatter(name="Container name", template="scenario_{scenario_id}_task_{task_id}")
container_id = create_container(image_name=image_name,
                                    container_name=container_name(scenario_id=scenario_id, task_id=task_id),
                                    command=command).set_upstream([download_video, fetch_prev_result])
started = start_container(container_id=container_id)

Hi, all. I am new to Prefect and trying to create a work flow with following need: 1. Get list of files from S3 2. Iterate the files and transpose the values in each file and dump to a db. I have following python functions: • function 1-> get S3 object • function 2-> read the S3 file content and transpose the value • function 3 -> write the transposed values to db What I am looking is how to link them in a flow so that function 1, 2, and 3 are executed for all the files I have on S3. How would I do it using Prefect? Thanks, Maz

Hi everyone! Is it possible to pass SQLAlchemy PG session through tasks? I've tried to create, but faced the pickling problems. Is there any predefined way to handle such stuff(sessions/temp objects)?

hi all, does Prefect Cloud have a way to view logs from flows, or is it recommended to integrate with another service for consuming log data?

Hi all! I have a state handler for alerting me when a

Task

is retrying. It tells me which

Task

failed, when it’ll retry, how many times this task has retried, etc. I think I’d really rather just have that same state handler on the top-level

Flow

to avoid setting that state handler on each task. I noticed the signatures for

Task

state handlers are different than those for the

Flow

state handlers — is there an easy way to have a

Flow

state handler that can access Task-level information? For example, given a

Flow

that’s changing states, can we readily access the specific

Task

in question?

Hello, recently, after I switched to M1 macbook, I found my prefect flow docker build and registry often frozen at this step below

- prefect.Docker | Pushing image to the registry...

Does it happen to other people?

Hi ! Background : I was beginning to write a few basic tests for my flows which basically extract data from db, transform and load data back into DB. Intention : I want to test my custom flow in term of : composition, serialization and tasks in terms of : high level structuring, input/output, mapping. Is there another place or point of reference for unit testing(think of CI) real time custom flows apart from link in this page (this places essentially points out how prefect unit tests its own platform if I am not wrong).

Hi guys! The question is about prefect/dask combo. Is there any way to limit dask scheduler not to push so many tasks on workers? In prefect cloud I've found Task Concurrency feature, but I believe this stuff will not work with dask. Thx)

Hey I might have missed something in the docs, so I’ll ask here: I have a kubernetesAgent which runs all my tasks, and we want to make stuff parallel. is it ok to use the mapping features (here) with a

LocalExecutor

? or is there something else I need to do? the docs here state that I need to a

DaskExecutor

for mapping to work…but I’m not sure which mapping, and I’m also not sure I understand why…

Hi all, Is there a documented recipe for deploying a Prefect flow on AWS ECS EC2? Specifically looking for a walkthrough for

run_config

parameters in

ECSRun

class

Hi all, I’m trying to get auto documentation using Sphinx from docstrings. The code base uses a lot of task decorators. I’ve done some research, but all those answers are ~ a year old, so not sure if things have not developed further. Ideally. I’d like to have a nice sphinx autogenerated documentation using `automodule`directive. Based on what I’ve found here and here. I can now get nice documentation, provided I define a function manually using

autofunction

directive (tedious and not ideal). I can achieve a similar behaviour when I define the task using class and run() method (would require a lot of changes in the code base, so not ideal too), and I’ve also came across a plugin for celery…. So my question is: is there anything else I can do with present version of prefect to set up sphinx documentation using

automodule

? If so, could you point me to the right direction/provide me with a code snippet? If not do you plan on making something like prefect-sphinx plugin, that would allow this functionality? Thanks in advance for any comment/suggestion!

Hi all, How do you authenticate to the Prefect GraphQL API using only Python's

requests

library? I need to cancel a bunch of flow runs but the Web interface is too cumbersome and the CLI tool doesn't support cancellation. I notice that the browser has a token that only lasts a few minutes.

Hi all, not a big deal, but my flows receive a label "Fields" out of nowhere. Is this expected? I don't have that label anywhere at the creation step....

Could you tell me how to invoke the class that is used between tasks? Thanks ya

Hello Everyone, I am really sorry I am asking this question in the wrong channel. I am trying to set up Prefect and using Bitbucket as storage.

STORAGE = Bitbucket(
        project="PrefectDemoECS",                            # name of project
        repo="prefect",                          # name of repo in project
        path="demo.py",                      # location of flow file in repo
        workspace="sandipviradiya",	
        cloud_username_secret="bb_username",
        cloud_app_password_secret="bb_app_password"
    )

It is showing me the 404 error as the given image. I checked the code https://github.com/PrefectHQ/prefect/blob/a2025983352d7e152da88358c3e578f3ce73778a/src/prefect/storage/bitbucket.py#L29 and it seems it is not selecting the Bitbucket cloud function and trying the Bitbucket server fetch only. Any guidance will be really helpful.

Thanks in advance everyone

Hi All, what is the best way to send email notification when flow is completed. Is it state handler?

Hi all, I wonder can someone help me make sense of some errors I am encountering with gcp auth. I am trying to store gcp credentials on the agent as described here so that my tasks are able to use google storage/big query. I am setting PREFECT__CONTEXT__SECRETS__GCP_CREDENTIALS on the agent via a helm chart to a string containing the json credentials. This seems to propagate GOOGLE_APPLICATION_CREDENTIALS to each prefect job, and the creds are different to what I set on the agent, but this var is set to the json contents rather than a path in the container containing the credentials. This causes errors for the prefect google utilities and the google api in python. I can hack a fix for this by running something like the following but I am wondering if this is expected behaviour or I am setting up the agent incorrectly?

from pathlib import Path
import tempfile
import os
import google.auth

creds = Path(tempfile.NamedTemporaryFile().name)
creds.write_text(os.environ["GOOGLE_APPLICATION_CREDENTIALS"])
os.environ['GOOGLE_APPLICATION_CREDENTIALS'] = str(creds)
google.auth.default()

Hello Everyone! I use

local agent

and try to bind local folder with folder inside docker container using

volumes

, but looks like it doesn’t work or I do it wrong 👀:

container_name = StringFormatter(name="Container name", template="scenario_{scenario_id}_task_{task_id}")
container_id = create_container(image_name=image_name,
                                container_name=container_name(scenario_id=scenario_id, task_id=task_id),
                                # command=command,
                                command=['ls'],
                                volumes=['data:/home/data',
                                         'temp:/home/temp'],
                                ).set_upstream(input_file_path)
started = start_container(container_id=container_id)
status_code = wait_on_container(container_id=container_id, upstream_tasks=[started])
logs = get_container_logs(container_id=container_id, upstream_tasks=[status_code])
log([logs])

The

log([logs])

shows, that the

data

and

temp

folders are missed. What am I missing here?

Hey everyone, I'm new to Prefect (and AWS) and have a quick question. When deploying your ECS Agent inside a EC2 in same AWS environment, do I still need to set authentication for the container deployments, and the S3 bucket with my scripts? Thanks!

Hey, how to do nested mapping?

hey everyone, I have a question about passing data between flows. My scenario is that I need to pass a pandas DataFrame from one flow to another. Right now I'm converting the DataFrame to JSON and passing it in as a Parameter. It works well, but I've only tried for relatively small datasets. I'm wondering: what are the limits of scale for this approach, and should I be handling this differently?

Hi all! Didn't find information if I can limit number of parallel jobs created by prefect agent (kubernetes agent). Can someone either throw me a link or answer. Thanks

Hello! I’m curious if in Prefect Server UI you can unarchive a flow? This would be the use case, I have multiple versions of the flow, the latest one is running, but I find there is a bug and I want to roll-back to the previous version. I know I can delete the latest version, but that doesn’t make the now latest version unarchived. Thank you!